Introduction
Hepatocellular carcinoma (HCC) is one of the most frequent tumours in the world - it rates as fifth with about 437.000 new cases every year [1]. The incidence of HCC varies considerably by country: it is below 5/100.000 in Caucasian populations in industrialized countries and it can reach 30-40/100.000 in some south-east Asian countries or in certain developing nations. Furthermore it has been proved that the incidence of HCC is increasing in many countries [2, 3-5]. The increase in the incidence of HCC is mainly linked with the increase in chronic hepatitis C [2-4, 6]. In France about 500.000 people seem to be affected by the hepatitis C virus [2]. It is estimated that about 2/3 of them present chronic hepatitis and that 20%, in absence of treatment, will develop into cirrhosis in 10–20 years. At this stage the incidence of HCC is between 1% - 4% a year [7]. It is estimated that in France the mortality rate associated with HCC arising from C virus cirrhosis will increase by 150% in men and 200% in women by 2020 [8]. It can thus be considered that in years to come HCC will be a problem to public health. The prognosis of this cancer is unfavourable. In fact only less than 25% of cases can be treated (liver transplant, resection, destruction with alcoholization or radiofrequencies), due to both operative contraindications (especially advanced cirrhosis) and locally advanced stages (multifocal lesions, invasion of the portal vein) or more rarely due to widespread metastasis [9, 10]. Palliative treatment can be proposed to patients who cannot undergo a radical cure and who have not reached a stage that is too advanced. Systemic chemotherapy achieves a low rate of response,without improving the rate of survival [10]. Non-invasive radiotherapy is of little use as it exposes the patient to a high risk of liver toxicity. Hence there are two main therapeutic approaches possible: chemoembolization with arterially injected Lipiodol - its efficacy remains controversial [11-14], and arterially injected metabolic radiotherapy. The concept of arterially injected metabolic radiotherapy is based on the liver’s double vascularization (hepatic artery, portal vein) and on the fact that HCCs are hypervascularized tumours which are mainly fed by the hepatic artery while 80% of the liver unaffected by the tumour is vascularized by the portal vein. Hence by injecting these therapeutic agents intra-arterially we can obtain a high “tumour/unaffected liver” ratio which enables us to administer a considerable tumour targeted dose, preserving the healthy part of the liver as long as possible. Various radioactive agents such as yttrium 90 labelled Lipiodol or yttrium 90 labelled microspheres can be used to perform arterially injected metabolic radiotherapy.

131iodine labelled Lipiodol
131iodine is a radioactive element that has interesting therapeutic properties: emission â (maximum 606 Kev), emission ã (principal, 365 Kev) which enable non-invasive research and hence a dosimetric approach. This emission ã also presents a drawback because it irradiates the environment too. The half-life of this radioactive element is also a disadvantage as concerns its duration (8 days). Lipiodol is an iodized oil initially used to diagnose HCC [15,16]. Lipiodol administered as a contrast medium by means of selective catheterization of the hepatic artery followed by a CT-scan enables the detection of HCC, or satellite tumours of previously diagnosed HCCs, which cannot be detected by conventional investigation methods (arteriography, CT-scan, echography) [16]. If on the one hand Lipiodol retention mechanisms in HCCs are not clealy explained, on the other hand certain factors have been recorded: the concentration of Lipiodol in peritumoural sinusoids [17], its fixation on cancer cell membranes and penetration into cancer cells [18,19]. The observation of prolonged Lipiodol retention in HCCs many years after its intra-arterial injection into the hepatic artery [18-19] has brought forth the idea of using this agent as a therapeutic vector.

Biodistribution of 131iodine. labelled Lipiodol
The first attempt at HCC labelling on man with 131iodine labelled Lipiodol dates back to 1986 [20]. The analyses of scintigraphic results focalized on the liver, lungs, stomach and thyroid for an 8-day period have highlighted a high “tumour/unaffected liver” ratio associated with a marginal activity in the mentioned organs, necessary conditions towards the performance of metabolic radiotherapy. Later the biodistribution of 131iodine labelled Lipiodol was studied on many patient groups [21-25], the largest of which were composed of 47 to 52 individuals [22, 25]. The main features of these biodistribution studies are summarized in table 1. Intra-arterially injected radiolabelled Lipiodol preferably fixes on the liver (but also on the lungs): the percentage of activity present in the liver varies from 70% to 90% with a tendency to decrease in time [21]. The ratio “tumour/unaffected liver” was high, 4.3±2.6, in the two largest groups, marking a significant increase in the course of time [22-24]. It has been stressed that the degree of Lipiodol fixation depends on the size of the tumour and it decreases if the size increases: 88% of HCCs<5cm present high retention (type 3 or 4 by Maki’s criteria [25]) while 74% of HCCs>10cm present low retention <50% (type 1 or 2 by Maki’s criteria). In one study [22], 14 patients did not present HCC but hepatic metastasis; the fixation ratio “metastasis/unaffected liver” was much lower than in HCCs (2.4±0.7 versus 4.3±2.6). One patient who presented a cholangiocarcinoma had a fixation ratio “metastasis/unaffected liver” that was almost zero[23]. The activity that could be identified in the lungs varied from 14% to 43% and increased in time. Significant fixations were not noticed in other organs. Lipiodol not fixed in the liver or in the lungs is eliminated mainly in the urine, 30%-50% of the activity injected at J8, but also marginally in the faeces, <3% at J5 [22]. The effective half-life of Lipiodol noticed in these studies is around 4-6 days [22, 23, 25]. In 1988 Nakajo [23] formulated a hypothesis, starting from a dosimetric study performed on this biodistribution data, that there could be a risk of radiation caused pneumopathy associated with the possible treatment of patients with pulmonary shunts. In fact the radioactivity injected to administer the tumour a dose of 100 Gy involved a lung dose of 39.2 Gy in 1 patient in 5 in his study. The lungs thus represent the critical organ in the realization of this treatment. If these biodistribution studies performed with small doses of radiolabelled Lipiodol, from 2 to 70MBq, have not observed a significant thyroid fixation in the scintigraphic results mainly performed at most 8 days after the injection, other more recent studies on the feasibility of this method, performed with high therapeutic activities (which border on 2GBq) followed by scintigraphic results (from 2 to 4 weeks after the injection) have highlighted a significant thyroid fixation in 7 patients on 12 [27] and in 13 patients on 13 [28].

Therapeutic use of 131iodine labelled Lipiodol
Many feasibility studies have been performed [27, 29-44]; their results are reported in tables 2, 3 and 4. The following observations can be made if we consider them all:

1. The 131I-Lipiodol administration method varies greatly as concerns the selective nature (proper hepatic artery) or hyper-selective nature (right and left hepatic arteries, even segmented) of the injections, the volume injected (from 2 to 20 ml) and especially the dose administered by the injection (74 MBq to 4.44GBq).

2. Studies state that a partial or incomplete response rate (over 50% reduction in size or in AFP dosage) can be noticed in 12% - 66% of patients; some hystologically documented cases of complete response are described [30,31,34].

3. The response rate increases with the decrease in HCC size and it is higher in nodular forms.

4. 131I-Lipiodol has a lasting antalgic action in patients affected by hyperalgic HCC.

5. The rate of survival at 6 months, 1 and 2 years varies respectively from 33% to 60%, from 12% to 50% and from 0 to 23%, as per studies.

6. Only one study highlighted dose/efficacy relations [37], probably due to inaccurate dosimetric calculations.

7. With selective injections, the reported tumour dosimetry is 34±32 to 64±54 Gy, the unaffected liver dosimetry is 3.3±1.5 to 5.5±8.7 Gy, and the lung dosimetry is 2.9±2.2 to 4.4±2.3 Gy. With hyperselective injections the tumour dosimetry is 37 to 183 Gy, the unaffected liver dosimetry is 2 to 15 Gy, and lastly, the lung dosimetry is 2.5 to 9.9 Gy.

Frequently noticed minor side-effects are asthenia, anorexia, temperature rise and increase in momentary pain; there can also be a passing alteration in liver values. Some rare, potentially serious side-effects can arise: serious asthenia, prolonged fever, haemorrhage in the digestive system, serious liver failure (4 deaths in one study comprising 26 patients [40]), and lastly pneumopathy, the cause of which is controversial - immunoallergic [35, 45] or exposure to radiations [46]. The frequency of these pneumopathies is hard to assess as in around 500 cases published, pulmonary complications have been reported in 10 patients [35, 45], 8 of which belong to the same group [46]. These pneumopathies arise gradually 4/6 weeks after the injection and lead to death in about 50% of cases. Little data is available concerning the dosage level received by the medical staff: only one study has given these results [47]. In this study, the nuclear physician was responsible for the injection and the pressure applied to the injection site. The dosage in the fingers is reported to be 19.5 mSv and the dosage in the entire body is around 110 mSv when the injection is given with a syringe. These values are very high and relevant compared to those we recorded in our ward (an average of 18.4 mSv in the chest and 285 mSv in the most exposed finger, unpublished personal data). An automatized injection system makes the injection dose fall on the fingers at less than 3 mSv and thus there remains only the dose bound to manual compression: 170 mSv in the fingers, 14 mSv in the entire body [47].

Only 3 randomized studies were performed [45, 48, 49]. In the first study, published in 1994 [45], 27 patients affected by HCC with portal vein thrombosis were randomly divided between treatment with 131I-Lipiodol, 2.2 GBq, (No=14) and supportive medical treatment (reference group, No=13). The intra-arterial injection of 131I-Lipiodol was repeated at 2, 5, 8 and 12 months. Its efficacy was assessed by the rate of survival, the progress in the dosage of alfafoetoprotein (AFP) and angiography. The two groups were comparable (classification by Child and Okuda, liver function, localization of the thrombus). This study has identified an important and significant improvement in the rate of survival (p<0.01) in patients treated with 131I-Lipiodol, with survival rates (IC 95%) at 3, 6 and 9 months which reached 71% (48%-95%), 48% (12%-55%), 7% (1%-31%) in the treated group versus the 10% (1%-33%), 0% and 0% in the reference group. The average survival period was 28 weeks in patients treated with 131I-Lipiodol versus 8 weeks in the reference group. This study was prematurely interrupted for ethical reasons due to the significance of the results. The tollerance to treatment was good except for frequent asthenia during the 15 days following the injection and one case of extensive interstitial pneumopathy of probably immunoallergic origin. The second randomized study was published in 1997 [48]. The patients who presented a non- resectable, non transplantable HCC were randomly divided between those treated with 131I-Lipiodol, 2.2 GBq, (No.=73) and those chemoembolized with 70 mg of cisplatinum (No.=69). The 131I-Lipiodol and chemoembolization injections were repeated, when possible, at 2, 5, 8, 12 and 18 months. The tumoural response was assessed by the size, the AFP dosage and the rate of survival; the latter was the main criterium in the assessment. 129 patients in total (65 in the 131I-Lipiodol group, 64 in the chemoembolized group) were assessed (excluding 13 patients, mainly due to the presence of portal vein thrombosis). No relevant differences were identified in the rate of survival: 69%, 38%, 22% 14% 10% at 6 months, 1, 2, 3 and 4 years in the 131I-Lipiodol group versus 66%, 42%, 22%, 3% and 0% in the chemoembolized group. Complete responses were respectively 1 and 0 in the 131I-Lipiodol group and the chemoembolized group, while partial responses were 15 and 16 (an insignificant difference). Tolerance was significantly better in the 131I-Lipiodol group, with 5 serious side-effects (1 pneumopathy, 1 liver failure, 1 extreme asthenia, 1 prolonged high temperature increase and 1 psychic stress and 0 deaths ascribable to the treatment) versus 29 serious side-effects (7 intense pain, 9 haemorrhages in the digestive tract, 9 liver failures, 2 cases of hemi-peritonaeum, 1 ischaemic cholecyst, 1 bronchospasm) and 6 deaths ascribable to the treatment, in the chemoembolized group. It was thus concluded that both therapeutic procedures were equally efficacious but the treatment with 131I-Lipiodol was remarkably better tolerated. The study of the third randomized group was published in 1999 and leaned towards 131I-Lipiodol as a post-surgical adjuvant treatment [49]. Forty-three patients initially treated surgically were randomly divided between the adjuvant treatment group with 131I-Lipiodol (1850 MBq, No=21) and the reference group, without adjuvant treatment, (No=22). In an average monitoring period of 34.6 months (from 14.1 to 69.7), six recurrences were recorded (28.5%) in the 131I-Lipiodol treated group versus 21 (59%) in the reference group (p=0.04). The average survival rates without recurrences both in the treated group and the reference one were respectively 57.2 months (0.4-69.7) and 13.6 months (2.1-68.3), (p=0.037). The survival rate at 3 years in the treated group and the reference one was respectively 86.4% and 43.6% (p=0.039). The study was prematurely interrupted due to the significance of these results.

Yttrium 90 labelled microspheres
Yttrium 90 is a radioactive element which presents satisfactory properties as concerns therapeutic purposes: pure emission â with half-life 64 hours, average energy 0,93 Mev (maximum 2,27 Mev) and an average tissue penetration of 2,5 mm (maximum 10,3). The absence of emissions ã complicates dosimetric studies which must turn to patterns and a pre-operative approach by means of a probe to identify the rays â. Many preliminary studies have used ceramic or resin (especially in liver metastasis) microspheres [50]. Though the rate of response was encouraging, some serious side-effects were highlighted, ranging from secondary medullary toxicity to the detachment of yttrium 90 from the microspheres, from post-hilus pulmonary fibrosis and secondary haemorrhages in the digestive tract to pulmonary and gastrointestinal shunts. The development of new microspheres in resin or in glass, stably labelled with yttrium 90, that is without systemic detachment of the radioactive element which is combined with the matrix of these microspheres, and the development of a dosimetric pattern [51, 52] assure greater safety to this therapeutic approach. This dosimetric model is based on the performance of liver scintigraphy after injecting macro-aggregates of 99mTc labelled human serum albumin (99mTc-MAA). It enables to study and quantify a possible pulmonary shunt, to detect a gastrointestinal shunt, to estimate the dosimetry in the tumour, the unaffected liver and the lungs. Seven feasibility studies have been published [53-59] since these two technological developments were introduced - (none randomized) involving a total of 163 patients with inoperable HCC. There were no cases of medullary toxicity. Investigations for a pulmonary shunt and the dosimetric approach to the lungs enable us to reliably identify and discard patients at risk of developing pneumopathy in the hilus pulmonis. In fact, in these studies only one patient presented a pneumopathy in the hilus pulmonis and he was rightly identified as a patient with a 39% risk of pulmonary shunt but he was treated all the same due to the rapid progress of his disease and the absence of every other treatment [59]. Despite the exclusion of patients with a gastrointestinal shunt detected by 99mTc-MAA scintigraphy, there were various cases of serious (or fatal) haemorrhages in the digestive tract: 1 patient on 10 in Shepherd’s study [54], 4 patients on 18 in Yan’s study [55], and 3 patients on 22 in Dancey’s study [59]. Various hypotheses were made to explain the onset of these haemorrhages in the digestive tract: the difficulty to distinguish a gastrointestinal shunt from hepatic activity [54], an increase in portal hypertension associated with the irradiation of the portal system, Yan [55], irradiation of the digestive tract due to a nearby hepatic lesion and lastly, a reflux of radiolabelled microspheres into the digestive tract [60]. The rate of response observed varies between 0 and 72% as concerns the size (reduction > 50% of the size) and between 30% and 100% as concerns AFP dosages (reduction > 50%), [54-57, 59]. The response rate (size) is linked with the estimated tumour dosimetry: in Lau’s study [56] 7/8 patients (87,5%) who had received more than 120 Gy in the tumour presented a partial response against 1/8 patients (12.5%) who had received a dose < 120 Gy. Two cases of complete histological response are described [57].

Conclusions and prospectives
We can say that the efficacy of 131I-Lipiodol treatment has been proved only in the treatment of HCC with portal vein thrombosis and as an adjuvant treatment which reduces the risk of recurrences in cases of operated HCC. This treatment is almost as efficacious and better tolerated than chemoembolization. As palliative treatment it achieves a partial response in 12% – 66% of cases. Furthermore 131I-Lipiodol has a lasting antalgic action in hyperalgic HCCs. Some minor side-effects are frequent while other rare ones can cause death (pneumopathies, liver failure). Lastly, protective measures against radiations (emission ã) require patients to be hospitalized and isolated in a protected room for a period of 8 days. This represents a drawback to the use of this treatment. New Lipiodol labelling methods are currently being studied [61], especially rhenium 188 which uses a less important component ã and which thus requires less severe radioprotective measures. In regards to the use of yttrium 90 labelled microspheres we can say that it is a potentially efficacious treatment, that the risk of pneumopathies in the hilus pulmonis calls for the performance of a scintigraph after an intra-arterial injection with 99mTc-MAA and that there exists a risk of serious haemorrhages in the digestive tract which 99mTc-MAA scintigraphy does not always foresee. However recurrences following this technique are less relevant than those following 131I-Lipiodol treatment. Lastly, the development of radiolabelled antibodies (anti-AFP antibodies or anti-ferritin antibodies) is another topic of research which will probably lead to the identification of new therapeutic agents to be used in intra-arterial metabolic radiotherapy [50].

E. Garin, P. Bourguet
Reparto di Medicina Nucleare
Centre Eugène Marquis
Rennes - Francia

Translation by Interpres SaS